🚀 go-pugleaf

How does `font harmonizing' software know what fonts are
========================================================
			just different `styles' of a particular `face'?
			===============================================

Utilities like Adobe Type Reunion on the Mac, and Font Minder in Windows,
etc group related fonts in font menus.  That is done so that `regular',
`bold', `italic', and `bolditalic' don't appear as separate entries
in font menus, for example.  Which is logical and shortens font menus.

Now, first of all, this seems somewhat superfluous, since Mac 
screen fonts and Windows PFM files can - and should - be set up to 
do the right thing in any case (although some font generation
utilities and font manipulation utilities don't know how). But, anyway...

In many cases `font harmonizing' utilities group fonts incorrectly.
For example, they may add fonts with related names to existing families
to which they do not belong.
Also these utilities can lead to missing fonts when font families are
`incomplete' - that is do not have all four of `regular', `italic', `bold',
`bolditalic'.
In addition, they get confused when a `family' has both an `italic'
and a `slanted' version, for example.

(*)	So, my question, is what information do these utilities base their 
	decision on?  And how do you stop them from doing this?

One conjecture is that fonts with the same FamilyName are grouped together.
But this is obviously a bad idea, since there are several groups of fonts
that have the same FamilyName, but are not simply different styles of the
same face.  For example, the three Adobe LucidaMath fonts all have the
same FamilyName, as do the four Y&Y LucidaNewMath fonts.  The five Adobe
Math Pi fonts all share the same FamilyName. And all 75 Computer Modern 
fonts have the same FamilyName! 

Next, one might image that all fonts with the same `root' FontName
should be grouped together, where the `root' is what you get after
stripping off `modifiers' like `Italic', `Kursiv', `Oblique', `Slanted',
`Bold', `Heavy', `Light', `Demibold', `Regular' etc.  But such a list of
modifiers is quite long, and somewhat arbitrary.  For example, some FontNames
use `Roman' to indicate the `regular' style, so this should be stripped off
too, but with other fonts it is part of the `root'.

Another conjecture is that grouping is based on font menu name (QuickDraw
name on the Mac and Windows Face Name in Windows).  But that can't be, 
since that is exactly the problem these utilities are designed to solve: 
namely that related fonts do not always have the same font menu name
(as they are supposed to).

Anyone have any solid information on this?  Or how to prevent specific fonts
from being `harmonized' that should not be?

Berthold K.P. Horn
Cambridge, Massachusetts, USA